Machine Learning Model for Lithographic Mask Pattern Generation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current lithographic technologies face challenges in accurately reproducing small feature sizes and high feature densities due to limitations in resolution enhancement techniques, leading to sub-optimal performance and increased errors in patterning processes.
Innovation Solution
A machine learning model is trained to generate characteristic patterns for mask patterns using a continuous transmission mask (CTM) and reference patterns, reducing metrics between the generated patterns and the CTM and reference patterns, thereby improving patterning process performance and manufacturability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If conventional resolution enhancement techniques are used, then lithographic patterning can be performed, but manufacturing precision deteriorates due to inability to accurately reproduce small feature sizes and high feature densities
Solution Approach 1:
The patent replaces conventional mechanical/optical resolution enhancement techniques with a machine learning-based computational system. The ML model processes CTM data to generate characteristic patterns, substituting physical resolution enhancement methods with an intelligent algorithmic approach that can accurately reproduce small features and high-density patterns without the limitations of traditional optical systems.
Solution Approach 2:
The patent transforms the continuous transmission mask (CTM) data through a machine learning model that learns optimal parameter transformations. The model adjusts pattern parameters such as feature sizes, spacing, and densities to generate characteristic patterns that meet manufacturing requirements, effectively changing the parameter space from raw CTM data to manufacturable mask patterns.
2Reliability
If conventional patterning methods are used, then manufacturing can proceed, but errors increase leading to sub-optimal performance
Solution Approach 1:
The patent implements a feedback mechanism where the machine learning model is trained using reference characteristic patterns and CTM data. The model learns from the relationship between input CTM patterns and desired output characteristic patterns, continuously improving its accuracy. This feedback loop enables the system to reduce patterning errors and achieve optimal performance by adjusting its internal parameters based on training data.
3Productivity
If traditional mask pattern generation is used, then manufacturing can continue, but critical dimension uniformity and yield are reduced
Solution Approach 1:
The patent performs preliminary pattern generation and optimization using the machine learning model before actual mask manufacturing. The model generates characteristic patterns that are pre-optimized for manufacturability, critical dimension uniformity, and yield. This preliminary computational action allows potential issues to be identified and corrected before physical manufacturing, improving both productivity and precision.
Data Source
AI summary
Methods of generating a characteristic pattern for a patterning process and training a machine learning model. A method of training a machine learning model configured to generate a characteristic pattern for a mask pattern includes obtaining (i) a reference characteristic pattern that meets a satisfactory threshold related to manufacturing of the mask pattern, and (ii) a continuous transmission mask (CTM) for use in generating the mask pattern; and training, based on the reference characteristic pattern and the CTM, the machine learning model such that a first metric between the characteristic pattern and the CTM, and a second metric between the characteristic pattern and the reference characteristic pattern is reduced.